Cutting mis-picks by 94 percent across three distribution sites

Three sites, one shared client base, and no shared view of stock. Pickers worked from printed lists that were already out of date by the time they reached the aisle, and every correction was found downstream by the packing team or, worse, by the customer.

Mis-picks were running at roughly one in every 120 lines. Each one cost the business twice: the credit note, and the goodwill. Peak season was eight weeks away and the operations manager had no confidence the existing process would hold.

We started by measuring rather than prescribing. Two weeks of observation across all three sites established where accuracy actually broke down, and it was not where the team expected: most errors happened at replenishment, not at pick.

ProVision WMS was configured for directed picking with scan verification at each location, and replenishment was moved to a task-driven queue rather than a visual judgement call. Handheld computers replaced the printed lists. Stock became visible across all three sites in one view, so a shortage at one location no longer meant a manual phone call.

Go live was staged one site at a time, with the smallest site first, so the process was proven before it reached the busiest floor.

Order accuracy settled at 99.7 percent within the first full month and has held there. Mis-picks fell by 94 percent against the pre-project baseline.

The less expected result was in cycle counting. With live inventory and scan verification, the team stopped needing a full count each quarter and moved to rolling counts, which returned several days of labour a year.